Benchmark Results: Desktop Usage, Less-Than-Ideal Conditions
We know how these cards perform under a typical benchmark. Now, let's turn our attention to other usage scenarios, mainly desktop-based workloads. Most apps outside of the gaming world really don’t make use of GPU-based acceleration. Those that do are typically graphics-related titles, along with a handful of video playback apps. Today, we're going to use a handful, including Adobe Photoshop CS4, Cinebench R11, and Cyberlink's PowerDVD 9 to play a H.264 video file. Let's see how they do in less-than-ideal, real-life usage conditions.
Adobe Photoshop CS4
Although we can’t generate performance figures using Photoshop, we can definitely say that the title only really needs an add-in graphics card capable of OpenGL support. GPU acceleration does work on the integrated Radeon HD 3300 processor, but stuttering was still apparent during zoom and rotate functions.
We were slightly surprised to see Photoshop CS4's GPU utilization. It occasionally rises above 50% during the test (particularly while zooming in on an image and rotating the view). This makes it an interesting benchmark, because we can see how these cards fare when we don't fully utilize all their available processing power. The graph above comes from measurement data taken with two successive, manual runs. As you can see, peak power consumption stays pretty much around the 140 W mark.
Here's the kicker. Since GPU-based performance is practically the same with most discrete graphics cards, the board with the lowest (peak) power consumption is the one you want to get for Photoshop. Of the graphics tested in this article, our winner is AMD’s Radeon HD 5670. The card manages to eat up an additional 14 W on top of the base system’s power consumption.